Paleomagnetic studies on Icelandic lavas in , and their relation to worldwide research in paleomagnetism

Transcription

1 Paleomagnetic studies on Icelandic lavas in , and their relation to worldwide research in paleomagnetism Leó Kristjánsson Emeritus Research Professor Institute of Earth Sciences University of Iceland 8th Nordic Paleomagnetism Workshop, Iceland 30 Sept 2017

2 In memory of Norman D. Watkins He was a very productive and generous colleague, who brought attention to Iceland s potential in paleo-geomagnetism He was often outspoken in his views on research, and he did not seek popularity

3 In memory of Norman D. Watkins He was a very productive and generous colleague, who brought attention to Iceland s potential in paleo-geomagnetism

4 A bit about myself: Born at Ísafjörður, NW-Iceland in B.Sc. Hons. Physics, University of Edinburgh M.Sc. Geophysics, Univ. of Newcastle on Tyne Ph.D. Geophysics, Memorial Univ., St. John s Involved in paleomagnetic projects since 1964, in as research scientist at the University of Iceland Other research interests: Magnetic surveys History of science Teaching: Undergraduate courses in mechanics, electromagnetism, geophysics, etc.

5 The main purpose of paleomagnetic studies so far in Iceland has been to assist in stratigraphic mapping of 1-16 M.y. old regions of the country Some advantages of the Icelandic lava sequences for this work: - Of the order of 10 6 lavas are accessible, of average thickness ~10 m - Complete exposures of 400 m or more in many mountainside profiles - Polarity zones averaging lavas - The polarity of the primary remanence can be estimated quite easily using portable fluxgate meters in the field - Several profiles may be combined into composite sections of many km in total thickness, using polarities and other stratigraphic markers Practical advantages in field work, compared to many other global regions: - Relative absence of soil & vegetation - No dangerous animals - No effects of chemical weathering - Few lightning strikes

6 Examples of profile locations Typical profiles

7 Sediment horizons and groups of similar flows are used in correlating between profiles

8 Occurrences of faulting are easily detected. Intrusions are rare and can be avoided.

9 Flow 32 In most of the profiles sampled to 2001, flows were numbered by the mappers

10 Numbering facilitated resampling of unstable flows, and sampling for other purposes

11 Eyjafjördur fjord, central N-Iceland Two composite sections through NW-Iceland, up to 40 km apart

12 Problems for paleomagnetic research here: - Formations <1 M.y. old are heterogeneous due to Ice-age influences, difficult to sample (tuffs, etc.) and often impossible to correlate - Too few radiometric dates are available - Geologists have shown little interest in the stratigraphy of the older regions of the country in recent decades, Many areas are still unexplored in this respect - Funding of projects has been quite limited - Weather conditions can be unstable - The field season in the northern part of the country and in highlands is only 2-3 months

13 Advantages of Icelandic lava series (1-16 M.y.) for paleogeomagnetic studies, a by-product of stratigraphic projects: a 95 J A/m - Little hydrothermal alteration (<100 C) in many areas - Only minor tectonic movements, tilts are often <8 - No magnetite anisotropy, nor any evidence of deformation - Viscous remanence is easily removed by AF at mt - The primary remanence is strong (avg. 4 A/m). Directions are consistent within each lava flow, k ~200 or more, cf. examples - Primary remanence directions in some profiles may show serial correlation, but the mean time between flows is 1000s of years Not needed: Extended AF - Thermal treatment - FORC - Multicomponent analysis - Unfolding - Intersecting great circles - Two-tier statistics - Bootstrapping -...

14 Significant contributions to paleomagnetism from Icelandic lavas : : Introduction of key concepts including Fisher statistics and virtual poles. Suggestions of a global timescale of geomagnetic reversals at irregular intervals 1954: First use of paleomagnetic polarities in local stratigraphic correlation 1956: First successful application of AF demagnetization to eliminate viscous remanence, thus improving within-unit consistency of primary field directions 1957: First record of a pole path during a geomagnetic polarity transition 1957: First observation of apparent weakening of the field intensity in transitions 1964: One of the earliest reports of short geomagnetic polarity events (subchrons) 1967: Publication of a composite polarity column from 1100 lava flows in E-Iceland 1968: Correlation of various magnetic properties in lavas with magnetite oxidation 1977: K-Ar dating of two subchrons in the Gilbert chron 1982: Analysis of 2163 reliable remanence directions, demonstrating N/R equality as regards number of flows, mean direction, a.s.d. and mean intensity

15 Paleomagnetic results from Iceland have not always been taken into account by international researchers. One example is in the development of a geomagnetic polarity time scale from 1968 onwards. On the left we have the polarities column deduced from thorough mapping and sampling of lavas in East Iceland. By 1968 it was clear that this column implied the occurrence of at least 65 reversals in the period 13 to 2 M.y. ago, i.e. 6 per M.y. on average in an interval of ~11 M.y. The 1968 Heirtzler et al. timescale on the right however indicated that less than 40 reversals had taken place in that interval. Dagley et al. 1967

16 The Heirtzler et al. scale was obtained by interpretation of magnetic anomaly lineations along certain ocean ridges. The rocks causing the anomalies were many kilometers below the magnetometers. Rapid upwards attenuation of short-wavelength features could be expected. Many major assumptions also had to be made about the processes of generation of the crust, its structure and subsequent tectonics, its magnetic properties, etc. Inevitably, the Heirtzler et al. scale was only a rough approximation to the actual timing of the Earth s polarity reversals. However, these deficiences did not deter a number of people from carrying out sophisticated statistical calculations on the time series of Heirtzler et al. Others related polarities in geological formations to this scale in detail. Later versions of the official time scale gradually acknowledged that it had missed a lot of reversals. In the meantime, I deduced already in 1982 from Icelandic results that the average rate of reversals was at least 8 per M.y.

17 Estimates of secular variation Measurements on cores of slowly accumulating marine sediments often indicated that the geomagnetic field direction did not vary much between abrupt reversals VGP latitude Ninkovich 1966 Parés & Lanci 2004

18 Most paleomagnetic studies on volcanics abroad in the 1960s and 1970s were small, so that low-latitude VGPs stood out. This resulted in erroneous notions: -That the low-latitude poles were due to some unusual temporary state of the Earth s core, different from the state during ordinary secular variation -That low-latitude poles should therefore be excluded from gsv computations, as for instance #21 in R. Doell s study on 32 Brunhes lavas in France. Its VGP was at 44 N, so Doell decided that the cutoff should be at 45! It was assumed that a mean pole coinciding with the geographic pole could be obtained from ~10 units. L.K., data from 1975 Doell 1970

19 Norman Watkins and myself argued against these notions, pointing out that mid- and low-latitude poles could be an important part of the secular variation Kristjánsson 1968 Watkins et al Kristjánsson and Jóhannesson 1999

20 Here you see that with groups of 30 successive lava flows in my partially ordered list of over 5200 flows, the a.s.d. of field directions can vary widely. The mean pole, even of 100 flows, can lie several degrees off the geographic one. Variations in a.s.d. values also occur between 300-lava groups, to some extent because their lavas are of different mean ages.

21 A map of over 5200 VGPs from Icelandic lava flows shows that a significant proportion lies in low latitudes Preferred longitude intervals for low-latitude VGPs are not evident

22 In fact, ~10% of observed Icelandic poles (from projects not targeting transitions or major excursions, ~evenly distributed 1-16 M.y.) are below 40 N or S latitude. That proportion increases with the age of lavas, see later. C.G. Harrison wisely suggested in 1980 treating the distribution as ~90% Fisherian and 10% uniform % - - The irrationality of excluding low-latitude VGPs from secular variation statistics can be illustrated by applying the same method to computation of the mean and standard deviation of ages of inhabitants in Iceland or such groups, cf. next slide

23 In order to apply the paleomagnetic method to finding the mean age of any group of Icelanders and its s.d., you would first discard everyone older than say 65, 67, or 69.3 years, depending on which cutoff value you favor

24 In some lava sequences, several successive units may be essentially coincident in time. See the 14 units on the right (taken from a paper on volcanics abroad), in view of the rule of thumb that the geomagnetic field direction changes by around 5 per century. In some published data sets, the number of independent directions is ~half that of units My general conclusion is that models of the geomagnetic field have often underestimated its variations, because the data sets from volcanic sequences on which they were based, - Included too few units - Covered too short intervals - Had been stripped of low-latitude VGPs - Sometimes contained much serial correlation

25 More on the geomagnetic secular variation: By 1987, I had acquired enough remanence direction data from Icelandic lavas, to be able to study if their angular standard deviation varied with time. of VGPs Measurements were made at 3 laboratories, totalling 3950 lavas Effects from alteration or tectonics do not cause this trend Kristjánsson and Jóhannesson 1989 My updated analyses used more data and stricter criteria, see next slide

26 As far as I know, comparable results are not available from anywhere else. I have not wished to speculate about the possibility of correlations with other properties of the field such as its intensity or the rate of reversals per M.y. Kristjánsson 2013

27 The high a.s.d. value at ~13 M.y. in the previous graph is partly due to an episode of geomagnetic instability, which I investigated further recently in two NW-Iceland fjords. The VGP was moving around irregularly in some 28 successive lava flows; scattered low- and mid-latitude poles (black) are also found in overlapping profiles. Kristjánsson 2016 The episode may have lasted of the order of 0.1 M.y. It will be useful in stratigraphic mapping, possibly reaching tens of kilometers along strike.

28 The excursion in NW- Iceland is somewhat similar to the Steens Mountain episode in Oregon (~17 M.y.) discovered by Norman Watkins in Paleomagnetism was haunted in by suggested ultrarapid variation of the magnetic field during a part of that episode. I was not surprised when this was finally determined to be an artifact caused by unstable remanence.

29 About the effects of alteration, as inferred from deep drilling in geothermal areas: Paleo-direction studies in Iceland have lately concentrated on unaltered flows and those in the upper part of the zone with smectite clays and zeolites such as chabazite and thomsonite. These minerals signify that temperatures during burial reached 100 C or less. There are indications that higher burial temperatures progressively cause decay of primary remanence and buildup of viscous remanence with high coercivity. In a 1900-m deep core hole drilled in E- Iceland in 1978, it was found that when epidote and prehnite have appeared with chlorite at temperatures ~250 C, the primary titanomagnetites are gone. They get replaced by secondary magnetite.

30 Points about the number of samples to be collected per unit Some people in paleomagnetism have been claiming that 5-10 samples should be collected from all igneous units, and AF-demagnetized up to 100 mt or more, even also thermally. Such requirements may apply to highly altered and old material. However, not to lavas in Iceland which yield very consistent remanence directions. In some of the oldest lava sequences of Iceland for instance, where 4 samples were collected per flow and demagnetized at 5 AF steps, it would have made no significant difference (<1 ) to profile-mean directions or their a.s.d. values, if only one sample had been sampled per flow and demagnetized only at 20 mt! Note that regardless of the number of samples measured from a lava flow and treatments applied, unavoidable influences on the lava-mean direction remain: -Uncertainty in the correction for tectonic rotation (including possible original tilt) -Local magnetic anomalies (1 m -10 km scales, 1000s nt) during emplacement -Undetected slight movement of outcrops caused by erosion processes -Magnetic field refraction?

31 My final topic concerns so-called absolute paleointensity determinations which have become a popular experimental subject with paleomagnetists. Instead of citing the complex literature on the background to such experiments, I will present my own plain views and some relevant evidence. -Remanent magnetism depends on delicate molecular interactions that can be disturbed irreversibly by even minor chemical changes and deformations. -The remanence will thus be affected in unpredictable ways by internal changes in the magnetic grains (of various compositions), as well as by external attacks to which the smallest grains and extremities of large grains are most vulnerable. -The spectrum of remanence as a function of blocking temperatures has been modified during the initial cooling of a lava. It is then further modified in situ by geological processes, and again by heating in the laboratory. - Hence it is quite possible that PI determinations will yield spurious results much more often than paleo-direction measurements. However, criteria for the reproducibility and within-unit consistency of absolute PI determinations seem by consensus to be much less stringent than those for remanence directions!

32 Criteria for acceptance of a PI determination should include: - Many samples from any unit must pass pre-tests - Variety in the samples, regarding for instance in their oxidation state - Close clustering of the results for each set of procedures applied - Close agreement of mean values from different procedures - Other checks on consistency where available, such as reasonable agreement between flow-mean intensities in rapidly erupted series But what is the actual situation in many of the recent publications? - Generally, a small proportion of samples pass pre-selection tests, often either none or (by chance?) just a single sample from a unit - Distances between samples, etc. tend to be poorly documented - Samples commonly yield results differing by a factor of 2 or more - Different procedures sometimes give incompatible outcomes - I have seen PI values changing much while directions stay constant - PIs from fresh volcanics <100 years old do not always agree with the in-situ field intensity Some examples of divergent results are given in the next slide...

33 Examples of PI values published in various papers include these, from which the respective authors have computed weighted averages: and 1.51 µt in the same unit 3.21 and , 36.1, 15.3, 14.9, 21.8, 24.6 and and , 7.52 and , 14.8, 5.64 and , 83.6, 79.7, 60.7 and , 52.4 and , 51, 42, 23, 35, 53, 32, 40, 45 and 42 34, 47, 43 and 25 Method A 17.1, 10.3, 12.0, 25.2, 16.6 and 16.6 Method C 58 Method A 46, 30, 40 Method B 25, 28, 42 Method C 30, 24, 40 A few years ago, I proposed two new methods to estimate absolute paleointensities...

34 Tauxe 2010 PINT06, 0-5 M.y. Note the very wide range of results at 60- odd degrees, from Iceland I estimated that ~2/3 of the PI values lie between 9 and 54 µt, i.e. within a factor of 6.

35 Kristjánsson 2013 Here we have the distribution of mean remanence intensities in 5200 lavas from Iceland. The five dark columns contain over 2/3 of that total number, within a range of factor 5.6. This amount of scatter is similar to that in the paleointensity values. But the remanence intensity in a lava depends on several variables besides the paleointensity!! Log scale

36 I leave you to think about the possible reasons for the scattered data in Tauxe s diagram. My conclusion is that the actual scatter in intensities of the field 1-16 M.y. ago in Iceland must be quite small most of the time. One contribution to it is due to a general decrease in mean intensity as the VGP moves to the Equator during reversals and major excursions. Wilson et al This was investigated in a semi-quantitative way by P. Dagley and R.L. Wilson in The graph is based on ~2600 lava samples from Iceland. The red lines which I have sketched, would Indicate that the virtual dipole moment falls off by a factor of ~4.

37 I have continued this kind of analyses since the late 1970s, using data from increasing numbers of lava flows and of samples/flow, with stricter criteria. Until ~2000, I was inclined to believe that the observed field intensity and/or the virtual dipole moment was more or less a linear function of VGP latitude. However, it gradually emerged that the curve levelled off at low VGP latitudes, perhaps corresponding to the flat tail of the frequency distribution of VGPs. The dots are geometric means of remanence intensities (10 mt) in 5200 flows. Kristjánsson 2013

38 A/m Kristjánsson 2008 Geometric averages of remanence intensities that would be measured at the VGP for each of 4970 flows. Standard errors are ~0.1 A/m The dipole moment does not decrease as much as the measured intensities, only by a factor of ~2.5 Tauxe 2010 A similar but less certain result can probably be derived from worldwide PI determinations

39 Summary and notes on paleomagnetic stratigraphy and rock magnetism in Iceland Paleomagnetic polarity zones in lava flows, observed in the field or by measuring oriented samples in laboratories, have played an essential role in research on the volcanic stratigraphy of Iceland since the early 1950s. Polarity transitions and major excursions have also been quite useful in some stratigraphic studies here. Great strides were made in In , sampling projects initiated by N.D. Watkins involved ~2500 lava flows and 180 K-Ar dates obtained by I. McDougall. After 1978, stratigraphic mapping in regions of Iceland >1 M.y. old has proceeded at a slower pace, largely due to scarce funds and lack of interest among geologists. Paleomagnetists from abroad have carried out several studies here since the late 1970s. However, they have mostly resampled previously sampled profiles rather than mapping new areas. Few radiometric dates have been published recently. Rock magnetism has not been very relevant to paleomagnetic research in Iceland. Some work on lava flows having general interest was carried out in the 1960s and 1970s, for instance regarding effects of oxidation and of hydrothermal alteration.

40 Summary and notes on paleo-geomagnetism in Iceland Research in the 1950s introduced some key concepts of paleo-geomagnetism, such as Fisher statistics, virtual poles, and AF demagnetization of VRM. Proof of the existence of intermediate directions led to general abandonment of self-reversal ideas. Results in the 1960s and early 1970s included finding of 65 reversals in an 11 M.y. time interval, as well as confirmation of subchrons and of large scatter of VGPs. Analysis of major collections published in the late 1970s and early 1980s proved that the distribution of VGPs was non-fisherian in latitude and uniform in longitude. Field properties are independent of polarity. At least 8 reversals occurred per M.y. Among later results: Solid evidence for a decrease in a.s.d. during the last 16 M.y. A quantitative estimate of the variation in mean field intensity with VGP latitude. Attempts to find absolute paleointensities from volcanics in Iceland and worldwide have often given doubtful results. PIs are less useful in geology than directions. Paleomagnetic field models will not be realistic unless proper account is taken of results from Iceland. Extending the research on remanence directions in Iceland and other prime-quality locations as was championed by Norman Watkins, would have brought us closer to understanding geomagnetism than is the case today.

41 Photo: Ágúst Guðmundsson There is still a lot of paleomagnetism to be done in Iceland... Thank you for your attention

42 Abstract of the guest lecture given at the 8th Nordic Paleomagnetism Workshop, held at the Leirubakki Hotel in S- Iceland, 30 Sept. to 7 Oct. 2017